Mailman 3 August 2006 - Twisted

[Twisted-Python] Twisted website
by Jean-Paul Calderone 03 Aug '06

03 Aug '06

The trac server has been modified to run under a somewhat different configuration. Please let me know if you notice any irregularities on the website. Jean-Paul

1 1

[Twisted-Python] Problem in the example code of Twisted Network Programming Essentials
by Mike Meng 03 Aug '06

03 Aug '06

Hi all, I'm learning Twisted by reading Abe Fettig's "Twisted Networking Programming Essentials". I tried the example 2-6 (dataforward.py) but the program hung after printing 'Connected to ...'. I don't know Twisted enough to figure out where the problem is. Will anybody tell me why? I paste the code here: -------------------------------------------------------------- from twisted.internet import stdio, reactor, protocol from twisted.protocols import basic import re class DataForwardingProtocol(protocol.Protocol): def __init__(self): self.output = None self.normalizeNewlines = False def dataReceived(self, data): if self.normalizeNewlines: data = re.sub(r"(\r\n|\n)", "\r\n", data) if self.output: self.output.write(data) class StdioProxyProtocol(DataForwardingProtocol): def connectionMade(self): inputForwarder = DataForwardingProtocol() inputForwarder.output = self.transport inputForwarder.normalizeNewlines = True stdioWrapper = stdio.StandardIO(inputForwarder) self.output = stdioWrapper print "Connected to server. Press ctrl-C to close connection." class StdioProxyFactory(protocol.ClientFactory): protocol = StdioProxyProtocol def clientConnectionLost(self, transport, reason): reactor.stop() def clientConnectionFailed(self, transport, reason): print reason.getErrorMessage() reactor.stop() if __name__ == '__main__': import sys if not len(sys.argv) == 3: print "Usage: %s host port" % __file__ sys.exit(1) reactor.connectTCP(sys.argv[1], int(sys.argv[2]), StdioProxyFactory()) reactor.run() ----------------------------------------------------------------------- Thanks in advance. Mike Aug. 3, 2006

1 0

[Twisted-Python] Writing benchmark applications with twisted - any examples?
by Erik Forsberg 03 Aug '06

03 Aug '06

Hi! I'm thinking about writing a benchmarking application using twisted. It needs to connect to a server via xmlrpc, and I'd like to simulate multiple clients connecting at the same time, so I need some concurrency. Each request needs to be unique (different usernames). Are there any examples of a twisted application that does this kind of thing? I guess a http benchmark could serve as a good example, if such a thing exists. Thanks, \EF -- Erik Forsberg OpenSource-based Thin Client Technology Systems Analyst/Developer Phone: +46-13-21 46 00 Cendio AB Web: http://www.cendio.com

1 0

[Twisted-Python] Restarting a reactor listener
by ian.parker＠facilita.co.uk 02 Aug '06

02 Aug '06

I have created an internet server that is started in a thread from a Python QT GUI. I can call reactor.stop() and sucessfully get notified of shutdown and the run thread terminates, the TCP/IP listen port is closed. When I restart it by calling reactor.listenTCP() and reactor.run(installSignalHandlers=0) a second time I find that things don't work correctly. e.g. TCP calls are accepted and my protocol.dataReceived() is called but then calling reactor.callInThread(self.blockingMethod, data) does nothing. Also calling stop() a second time does not stop the running reactor thread or terminate the TCP listener. I have tried numerous options, and have been careful to avoid any conflicts with the GUI threads. It appears that the reactor is not being restarted correctly after stop() and a second run(). What I need is to stop and re-start the listenTCP. Is there a way to do this? Thanks, Ian

4 4

[Twisted-Python] [OT] Job offer: Flumotion
by Thomas Vander Stichele 02 Aug '06

02 Aug '06

Hey everyone, Fluendo is looking for a developer to join their team of streaming server developers and expand the video-on-demand functionality of Flumotion. Tasks include integration of the functionality in the streaming server core, the cluster platform, and providing a web-based frontend for managing and transcoding customer's VOD files. Experience with at least one of multimedia, Linux/UNIX, Python, Twisted, Web service programming is a definite plus. Applicants are expected to be willing to relocate to Barcelona, Spain. Fluendo offers a competitive salary, private health care, and an attractive bonus system based on performance of the developed platform. For more information, see http://www.fluendo.com/ Interested developers can send their application and resumé to me at thomas <at> fluendo <dot> com Thanks, Thomas

1 0

[Twisted-Python] "Injecting" a callWhenRunning when the reactor is already running?
by justind2＠ussonet.net 02 Aug '06

02 Aug '06

Here is a script I've made and am playing with to try to understand how things behave in Twisted. I'm using twistedsnmp but I think my problem here is my understanding of how to use the reactor (or lack there of). I feel like I'm misunderstanding some fundamentals, but I may just be hard-headed. I'm not even sure if this makes sense. I've noticed that unless reactor.callWhenRunning(return_def,df) is ran prior to calling reactor.run() the return_def will never be ran. I'm not really sure how to phrase the question, because it may be the wrong one.. so here goes... Is there a way to get the reactor to acknowledge a new "callWhenRunning" after the reactor.run()? I've included my bad code and working code, the difference is in the "bad code" I launch a thread that tries every 5 seconds (after reactor.run()) to do what the "working code" does only once before. Any help would be appreciated. BAD CODE BELOW: from twisted.internet import reactor from twisted.internet import task from twistedsnmp import snmpprotocol, agentproxy from twisted.internet.defer import TimeoutError import pprint import logging import thread import time logging.getLogger().setLevel(logging.INFO) ips = ['10.0.16.2','10.0.16.18','10.0.16.34','10.0.16.66','10.0.16.82','10.0.16.98','10.0.16.114','10.0.16.130'] devices = {} logging.info('Building device dictionary') for ip in ips: devices[ip] = {} devices[ip]['oids_to_poll'] = {} devices[ip]['oids_to_poll']['sysDescr'] = '.1.3.6.1.2.1.1.0' devices[ip]['oids_to_poll']['uptime'] = '.1.3.6.1.2.1.1.3.0' logging.info('Device dictionary built') def successful_response(result): logging.info(str(result)) return result def errored_response(err): if err.trap(TimeoutError): logging.error("timeout error") #### dont return the error because it will be handeled if you do else: logging.error("error" + str(err)) return err def return_def(df): return df def process_devices(): logging.info("proccess_devices thread has been launched") while True: time.sleep(5) for device in devices.keys(): #### first build the proxy try: logging.info("building proxy for device:" + str(device)) port = snmpprotocol.port() proxy = agentproxy.AgentProxy(device, 161, community='public',snmpVersion='v2',protocol=port.protocol,) except: logging.exception("exception occured while building proxy") else: ##### now that the proxy is buil send the request for the oids to poll if devices.has_key(device): try: oid_descriptions = devices[device]['oids_to_poll'].keys() except: pass else: try: oids = [] for oid_description in oid_descriptions: oids.append(devices[device]['oids_to_poll'][oid_description]) except: pass else: ##### you have built the list of oids to poll now build the deffered and return it logging.info("building defferred for device:" + str(device)) df = proxy.get(oids, timeout=0.25, retryCount=3) df.addCallback(successful_response) df.addErrback(errored_response) logging.info("launching callWhenRunning for device:" + str(device)) reactor.callWhenRunning(return_def,df) thread.start_new(process_devices,()) logging.info("starting reactor") reactor.run() WORKING CODE (not necessarily good): from twisted.internet import reactor from twisted.internet import task from twistedsnmp import snmpprotocol, agentproxy from twisted.internet.defer import TimeoutError import pprint import logging import time logging.getLogger().setLevel(logging.INFO) ips = ['10.0.16.2','10.0.16.18','10.0.16.34','10.0.16.66','10.0.16.82','10.0.16.98','10.0.16.114','10.0.16.130'] devices = {} logging.info('Building device dictionary') for ip in ips: devices[ip] = {} devices[ip]['oids_to_poll'] = {} devices[ip]['oids_to_poll']['sysDescr'] = '.1.3.6.1.2.1.1.0' devices[ip]['oids_to_poll']['uptime'] = '.1.3.6.1.2.1.1.3.0' logging.info('Device dictionary built') def successful_response(result): logging.info(str(result)) return result def errored_response(err): if err.trap(TimeoutError): logging.error("timeout error") #### dont return the error because it will be handeled if you do else: logging.error("error" + str(err)) return err def return_def(df): return df def process_devices(): logging.info("proccess_devices thread has been launched") for device in devices.keys(): #### first build the proxy try: logging.info("building proxy for device:" + str(device)) port = snmpprotocol.port() proxy = agentproxy.AgentProxy(device, 161, community='public',snmpVersion='v2',protocol=port.protocol,) except: logging.exception("exception occured while building proxy") else: ##### now that the proxy is buil send the request for the oids to poll if devices.has_key(device): try: oid_descriptions = devices[device]['oids_to_poll'].keys() except: pass else: try: oids = [] for oid_description in oid_descriptions: oids.append(devices[device]['oids_to_poll'][oid_description]) except: pass else: ##### you have built the list of oids to poll now build the deffered and return it logging.info("building defferred for device:" + str(device)) df = proxy.get(oids, timeout=0.25, retryCount=3) df.addCallback(successful_response) df.addErrback(errored_response) logging.info("launching callWhenRunning for device:" + str(device)) reactor.callWhenRunning(return_def,df) process_devices() logging.info("starting reactor") reactor.run()

5 7

[Twisted-Python] How to tell when ReconnectingClientFactory gives up
by James Bunton 01 Aug '06

01 Aug '06

Hi, Is there any way to tell when a ReconnectingClientFactory has stopped trying? I'm talking about due to connection failures, etc As far as I can tell, in ReconnectingClientFactory.retry, if retries > maxRetries, then the function exits with a message to the log. I can't find a good way to trap this event. The only things I can think of are to either hook clientConnectionLost and clientConnectionFailed and count when we've hit maxRetries, then trigger whatever I want to do, or to patch ReconnectingClientFactory to do the same. Anybody got any ideas? --- James

1 0

[Twisted-Python] IOCP reactor
by Jean-Paul Calderone 31 Jul '06

31 Jul '06

IOCP reactor has been offline for almost a month now. Hey, Windows developers, think we can do something about this?

1 0